Security & Trust

Trust, built in - not bolted on.

You are putting your brand, your content, and your learners in our hands. Here is exactly how we keep them isolated, verifiable, and under your control at every step.

The foundations

Six things we get right by design

🏢

Tenant isolation

Every workspace, template, asset, job, version, and report carries an organization identifier. Nothing is readable or mutable across tenants.

🧑‍⚖️

Human in the loop

The AI drafts; you decide. Each stage requires your approval, and nothing exports until you sign off.

🧾

Signed provenance

Every exported deck embeds its template version, content digest, and a validation report - so you can always prove what shipped.

🔐

Encryption & secrets

Content and credentials are encrypted, and connector secrets live in a managed secrets service - never in plaintext or the browser.

📋

Audit trail

Authentication, approvals, exports, and administrative actions emit immutable audit records with actor, action, and timestamp.

🔑

Deny-by-default access

Role-based access control with least privilege. Every read, export, and delete is authorized against the object's tenant and role.

Clear lines

What we never do

Never share content across customersYour courses, templates, and data are yours alone.

Never store credentials in plaintextNo tokens or passwords in the browser or application records.

Never export unverified outputA deck that fails any validation or per-slide check is withheld, not shipped.

Never run untrusted uploaded codeUploaded templates produce data-only analysis, never executable plugins.

By construction

What always holds

✓

Brand-locked exportDecks start from your exact approved template and are byte-verified against it.

✓

Every slide is checkedA per-slide screenshot gate confirms design, header, bullets, and icon before you see it.

✓

Fail-closed by defaultWhen any safety or quality check cannot pass, the system withholds output and tells you why.

✓

You own your exportsDownload your assets with their provenance and validation evidence, anytime.

Responsible AI

Assistance you can defend to a reviewer

Generated content passes a panel of senior-reviewer AI agents graded on teaching quality and platform standards, and is held below the acceptance bar while any required-fix issue is open. You remain the author of record: you review, edit, and approve every asset. We surface what was generated and how it was checked, so your name only goes on work you trust.

Security questions before you buy?

Bring your checklist - isolation, provenance, retention, access. We built the platform to answer it.

✦ Open the studio